[tei-council] FAND faces setback

Thu Feb 23 16:25:28 EST 2006

> FWIW, there is no XSD-generated code per se, its all done with
> trang.

I know. The special-casing might have to be hacking up the RelaxNG
before it's handed to trang for XSD conversion.

>   1. pure FAND: kill numbered divs now and forever. anyone who wants 
>      them back writes their own content model

>   2. loosen the current content model to allow interleaved <div>,
>      <div1> and <div0>. it would be nonsense, but so? does it really
>      justify immense mental effort and time spent protecting people 
>      from being dumb?

Perhaps not *this* particular expenditure, but yes, in the overall
picture our mental effort to protect people from being dumb is effort
well spent.

>   2a. do 2, but add a schematron rule to say you can't have div and 
>       div1|div0 in the same [<front>, <body>, or <back>]

This might be a good solution.

>   3. trap the situation in Roma code ad hoc

I don't understand where & how you're talking about trapping, but
since you "really really don't want to do" it, perhaps it doesn't
matter.

4. Fix odd2odd so that DTD generation is non-deterministic, and tell
   XSD users "can't do that". (Alright, probably not a good idea,
   esp. since numbered-div-huggers are often XSD users.) Sigh.